Walkability assessment of metro catchment area: A machine learning method based on the fusion of subject-objective perspectives

Abstract

China's metro system is developing rapidly. Walking is the most frequently adopted mode to connect to the metro, the attributes of the pedestrian-built environment around the stations directly influence people's willingness to use the metro. However, few studies have paid attention to the comprehensive assessments of the built environment in the metro catchment area. Thus, this paper attempts to construct a walkability evaluation model that combines subjective and objective perspectives. We collected field data of the built environment factors affecting on walkability in the 800 m buffer zone of eight case metro stations in Dalian city, China. We also collected on-site interviews from 867 passengers to evaluate the walkability. A machine learning-based approach was developed to calculate the weights of walkability variables, followed by constructing a Score-Effectiveness framework to identify the built environment factors in the metro catchment area that need to be improved. The study found that the shading facilities, obstacle barriers, and resting seats around pedestrian walkways are the most efficient and imbalanced variables recognized by the crowd. The convenience of overpasses and underpasses are additional efficient imbalance-type variables for leisure and commuting populations, respectively. This indicates that the current level of construction of the above five built environment factors is relatively low, but the construction has a significant impact on the degree of friendliness in supporting pedestrian walkability. In this paper, improvement measures are proposed in a targeted manner in order to achieve the effect of effectively improving the current level of metro catchment area's walkability. The results of the study can provide references to provide strategies for precise pedestrian planning in the metro catchment area, leading to a pedestrian environment with high walking quality.

Keywords: built environment; efficient imbalance-type factors; machine learning; metro catchment area; optimization strategies; walkability.